Abstract
An interim analysis procedure that relies on combining the p-values from tests of significance on separate segments of a preplanned total sample was evaluated with respect to contingent increase in that sample size following an interim look at the data. The use of the interim test criteria to identify studies that are neither already too promising nor too negative to profit from increased sample size was examined. Type I error probabilities were not adversely affected by the contingent increase in sample size, but the power enhancement was achieved at cost of a commensurate increase in expected sample size. Nevertheless, if one proposes to use the interim results as basis for a contingent increase in the total sample size according to any of the plans that were evaluated, there would seem to be no statistical basis for objection.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call